This thesis aims to analyze biomedical images such as X-rays, CT scans towards assistance to the diagnosis of chest diseases. In particular, it will consist of creating image analysis approaches which exploits machine learning principles for identifying types of respiratory disease, localizing and evaluating the severity of infections and generating relevant health indicators.
Contact : karim.hammoudi@uha.fr
PhD thesis:
Generalizability of artificial intelligence models for chest-related multi-disease diagnosis assistance from bioimagery
Start. date: October 2024
Context: Since 2020,research works in biomedical imagery have been intensified for analyzing chest-related diseases, notably in reason of pneumonia caused during the COVID-19 pandemy. In this context, our team, which investigates the area of image recognition from Convolution Neural Networks (CNNs) since 2016, contributed to the worldwide efforts towards designing efficient pneumonia detection models by exploiting chest X-ray images.
Background: Our first investigations [1] were focused on exploiting our image classification skills and approaches for training various deep learning models over publicly available chest X-rays (normal and pneumonia cases). Although the quantity of labeled COVID-19 X-rays were initially limited for the community, we tailored efficient pneumonia classification models.
(c) GdR IASIS - CNRS - 2024.